public inbox for xenomai@lists.linux.dev
 help / color / mirror / Atom feed
From: Philippe Gerum <rpm@xenomai.org>
To: Florian Bezdeka <florian.bezdeka@siemens.com>
Cc: xenomai@lists.linux.dev,  Jan Kiszka <jan.kiszka@siemens.com>
Subject: Re: [PATCH RFC Dovetail 6.16 2/2] genirq: irq_pipeline: Migrate mask_lock to a hybrid spinlock
Date: Fri, 26 Sep 2025 09:23:49 +0200	[thread overview]
Message-ID: <875xd5ismy.fsf@xenomai.org> (raw)
In-Reply-To: <20250924-wip-flo-fix-lockdep-warning-for-6-16-v1-2-8a8475e2cdc0@siemens.com> (Florian Bezdeka's message of "Wed, 24 Sep 2025 23:23:15 +0200")

Florian Bezdeka <florian.bezdeka@siemens.com> writes:

> The reason for the migration is the lockdep report below.
>
> The irq_desc_lock_class is the dovetail specific hybrid spinlock.
> pci_config_lock and mask_lock are both raw spinlocks.
>
> The behaviour of the pci_config_lock is different on kernels with
> IRQ pipelining enabled. While on plain Linux the critical section
> is entered with hard IRQs disabled, the inband stage is stalled with
> hard IRQs enabled on CONFIG_IRQ_PIPELINE. This is the reason why the
> pipeline entry code may still run on the same CPU and grab a
> (hybrid) IRQ descriptor lock.
>
> The critical section protected by the mask_lock was always entered
> with hard IRQs on.
>
> To break the cycle we migrate the mask_lock to a hybrid spinlock.
> This lock is only grabbed for IRQ setup. It's expected to have the
> lowest RT impact.
>
> irq_setup_affinity() is expected to be called from the inband stage
> only, so we will now enter the critical section with hard IRQs
> disabled. With that the pipeline entry code is locked out, breaking
> the cyclic lock dependency.
>

LGTM.

-- 
Philippe.

  reply	other threads:[~2025-09-26  7:34 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-09-24 21:23 [PATCH RFC Dovetail 6.16 0/2] Dovetail: Fix lockdep warning observed on 6.16 Florian Bezdeka
2025-09-24 21:23 ` [PATCH RFC Dovetail 6.16 1/2] locking: irq_pipeline: Rename MUTABLE_SPINLOCK to HYBRID_SPINLOCK Florian Bezdeka
2025-09-24 21:23 ` [PATCH RFC Dovetail 6.16 2/2] genirq: irq_pipeline: Migrate mask_lock to a hybrid spinlock Florian Bezdeka
2025-09-26  7:23   ` Philippe Gerum [this message]
2025-09-28 12:24 ` [PATCH RFC Dovetail 6.16 0/2] Dovetail: Fix lockdep warning observed on 6.16 Philippe Gerum

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=875xd5ismy.fsf@xenomai.org \
    --to=rpm@xenomai.org \
    --cc=florian.bezdeka@siemens.com \
    --cc=jan.kiszka@siemens.com \
    --cc=xenomai@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox